|Home | About | Journals | Submit | Contact Us | Français|
Although the statins have effectively ended the debate on the benefit of cholesterol lowering, there is less agreement on how and when they should be used in the primary prevention of vascular disease. Some investigators favour aggressive intervention, on the premise that endemic diseases need fundamental action;1 others take a more conservative stance, fearful that the high prevalence of occlusive arterial disease a wholesale attempt to counter would overwhelm an already stressed healthcare system.2
Geoffrey Rose,3 the doyen of preventive cardiology, stated in 1991 that 'all policy decisions should be based on absolute measures of risk; relative risk is strictly for researchers only'. By extrapolation then, risk management for prevention of coronary heart disease should be based on absolute global risk assessment, so that scarce healthcare resources can be targeted at those patients in greatest need. Remarkably, within the space of a decade, this advice has been enshrined in two series of guidelines providing recommendations for prevention of coronary heart disease in clinical practice. In 1994, a Task Force with representation from the European Atherosclerosis Society, the European Society of Cardiology and the European Society of Hypertension emphasized the importance of overall vascular risk assessment in a report4 that unified preventive strategies for vascular disease across Europe. This document has recently been updated and modified5 to make it applicable to more than thirty European countries.6
Enlightenment in Europe has spread to the USA. The third report of the Expert Panel on Detection, Evaluation and Treatment of High Blood Cholesterol in Adults—a group within the framework of the National Cholesterol Education Program that has driven national policy on cholesterol management since the 1980s—although still focusing on LDL cholesterol as the primary target of therapy, expands its horizons to include other lipid and non-lipid risk factors within a management portfolio for prevention of vascular disease based on the absolute risk of the patient.7 The principles of the European and US coronary prevention guidelines are therefore fundamentally the same, promoting the use of global risk assessment and treatment strategies that broaden choice for the clinician and expand options for the citizen.
In both Europe and America there is strong emphasis on the importance of achieving targets, and in Europe these are a total cholesterol of 5 mmol/L and a low density lipoprotein (LDL) cholesterol of 3 mmol/L, respectively. However, we need more assurance that treatment to these target values makes best use of limited healthcare resources. The curvilinear relation8 between cholesterol concentrations in plasma and risk of coronary death is well established but poorly appreciated. If this relation is maintained during cholesterol-lowering treatment, then a 1.3 mmol/L fall in cholesterol, from 7.8 to 6.5, should produce a substantially greater reduction in events than would a similar fall from 6.5 to 5.2 (Figure 1). This analysis provides a credible explanation for the assertions of the investigators from the West of Scotland (WOSCOPS)9 and Cholesterol and Recurrent Events10 studies that, as cholesterol concentrations in plasma are lowered further by statin treatment, vascular risk reduction becomes more and more attenuated. By implication then, strenuous efforts to drive cholesterol concentrations further downwards could be excessively zealous, since for most mildly hypercholesterolaemic individuals a reduction in this lipid fraction of about 25% appears to yield all the benefit expected from intervention.9
The economic importance of the argument becomes even clearer when we remember11 that doubling the dose of a statin will only reduce plasma cholesterol by a further 5-6%. That is, two-thirds of the maximum lipid-lowering benefit of a statin is realized at the starting dose. So, clinicians face several conflicting dilemmas in their management of hypercholesterolaemia to prevent first heart attacks. On the positive side, they know that statin treatment is effective in reduction of vascular events even in patients with only modestly raised coronary risk. However, they are under cost constraints, and have to focus their attention on individuals with the greatest need. In the UK, this group comprises individuals whose global risk is greater than 3% per year.12,13 (Choice of this 3% figure was determined partly by the high cost of statins, and with the advent of cheaper generic preparations the picture will doubtless change; the present argument is based on current prescribing costs.) Such individuals usually harbour several risk factors and are frequently severely hypercholesterolaemic. So treatment guidelines all specify target lipid concentrations that will sometimes be difficult to achieve in high-risk patients. Depending on the price structure of the drug, doubling of the statin dose could double the cost of managing these high-risk individuals, if doctors are constrained to treat to a specific cholesterol target.
Behind all of the above there is a disquieting sense that chosen target concentrations may not have a sound scientific basis, and that there is little to be gained from attempts to reduce cholesterol concentrations (in patients other than those with familial hypercholesterolaemia) by more than 20-25%.9 The experience of WOSCOPS participants (men with a raised cholesterol but no history of myocardial infarction), offer clues on how to proceed. At baseline, their average LDL cholesterol concentration was 4.96 mmol/L and, on the assumption that treatment for all was an appropriate option, the European guideline target for these individuals was 3.0 mmol/L. What actually happened (intention-to-treat analysis) was that only one-fifth of the pravastatin treated group reached this target (Figure 2). The average LDL cholesterol in the statin treated cohort was 3.67 mmol/L, and individuals in the highest LDL quintile (whose adherence to the treatment has to be considered very doubtful) showed little improvement from their baseline lipid values.
The treatment strategy in WOSCOPS was to use the maximum recommended dose of pravastatin (40 mg/day). Greater LDL reduction would have needed a change to a more powerful statin than pravastatin, say atorvastatin, of which 10 mg/day achieves about the same reduction in LDL cholesterol as 40 mg pravastatin.11 If we then assume that the LDL response curve to 10 mg atorvastatin will be the same in practice (i.e. including those who do not adhere to treatment) as that to pravastatin in WOSCOPS, and that further reductions of 6% were obtained every time the atorvastatin dose was doubled (Figure 3), then even at the highest (80 mg) atorvastatin dose only about 60% of the WOSCOPS cohort would have reached European guideline targets.
On the basis of these considerations, does statin deployment which targets a limited number of individuals at highest risk and excludes the majority of the population make best use of limited resources? Consider two alternative strategies, which we will call treat-to-target (Box 1) and fire-and-forget (Box 2). ('Forget' is not, of course, to be taken literally.)
Using fire-and-forget it is possible, for the same cost as treat-to-target to avoid twice as many vascular events in lower risk individuals by prescribing the starting dose of the drug more widely and resisting the temptation to drive the LDL cholesterol concentrations of that small group of patients at the highest level of risk downwards to target values. There might also be a safety benefit. Although statins are among the safest drugs used in clinical practice, experiences with cerivastatin remind us that, at high doses and in certain coprescription combinations, they can lead to severe, even life-threatening complications. Even though all available data from clinical trials show that benefit in prevention of cardiovascular disease far outweighs risks of myopathy, which the more hydrophobic, cytochrome P450-metabolized reductase inhibitors seem to possess, safety considerations favour spreading low-dose therapy more widely rather than attempts to achieve target concentrations whose clinical validity still awaits confirmation from continuing trials such as Treat to New Targets and SEARCH.
Box 1 The treat-to-target strategy
• Treat 10 000 middle-aged WOSCOPS-equivalent individuals with atorvastatin to reach European LDL cholesterol guideline targets
• From Figure 3, even with good adherence to treatment:
20% would need 10 mg of atorvastatin/day
20% would need 20 mg of atorvastatin/day
60% would need 40 mg of atorvastatin/day.
• Total daily drug outlay would be 30 000 mg
• If we ascribe to the group a global coronary risk of 3% per annum (equivalent to the limit set by the Scottish Intercollegiate Guidelines Network [Ref 13]), the number of vascular events avoided over 5 years will be — according to the statin trial evidence — one-third of 1500 or 500 events.
Eight large randomized controlled trials have established that statins can prevent first14,15 or recurrent vascular events16-19 in individuals with or without a history of coronary artery disease or with a mixed clinical background.20,21 Their remarkable concordance in lowering coronary risk by 30-40% over 5 years of treatment, despite the deployment of four different drugs (lovastatin, pravastatin, simvastatin, fluvastatin) suggests that the benefit is a feature of all statin therapy and is not agent-specific.
Box 2 The fire-and-forget approach
• Deployment of atorvastatin at a dose of 10 mg/recipient/day will treat 30 000 similar individuals for 5 years
• Only 20% would reach guideline targets
• Assume, pessimistically, that for the others the risk reduction achieved with the 10 mg dose changes pro rata with LDL response to treatment
• Then, the risk reduction achieved by 20% of the cohort will be in line with the statin trials — i.e. one-third — although, in a worst-case scenario, the reduction in the remaining 80% would be at least 20%
• The assumptions would then result in avoidance of 300 events in 5 years in 6000 patients who achieved European guideline targets and 720 events in 24 000 individuals whose reduction in LDL cholesterol was more modest. The total number of events avoided for the same expenditure would be 1020, or more than twice the number who benefited by treating to target.
The Anglo-Scandinavian Cardiac Outcomes Trial—Lipid Lowering Arm (ASCOT-LLA) compared the effect of atorvastatin (10 mg/day) with placebo in normolipidaemic hypertensive patients who had not experienced a coronary event but had at least three additional cardiovascular risk factors and were receiving concomitant blood pressure regulating therapy.22 The study was stopped early, after a median follow-up of 3.3 years, on a recommendation from its data and safety monitoring board based on overwhelming evidence of benefit in the primary endpoint (coronary heart disease death plus nonfatal myocardial infarction) and a small reduction in stroke. In ASCOT, the difference in total cholesterol between the treatment groups at the end of the study was 1.3 mmol/L, and because of their lower baseline total cholesterol values (at baseline, total and LDL cholesterol were 5.48 and 3.44 mmol/L, respectively), about 70% of the atorvastatin treated group reached the recommended European guideline target values. This trial therefore represents the treat-to-target principle.
A total of 6595 men, aged 45-64 years, were randomized in WOSCOPS and studied over an average of 4.9 years. Baseline plasma and LDL cholesterol levels were 7.03 mmol/L and 4.96 mmol/L. Participants received 40 mg pravastatin or placebo each evening. There was no significant difference between the treatment groups in their rates of withdrawal from study drug at any time during the trial. The drug, on the intention-to-treat principle, lowered plasma cholesterol by 1.0 mmol/L. So, as noted above, and because of their higher baseline cholesterol values, only about 20% of the WOSCOPS pravastatin cohort reached European target cholesterol values. WOSCOPS therefore represents the fire-and-forget approach.
Both ASCOT and WOSCOPS, though differing in detail, were broadly similar in design (Table 1). It is therefore reasonable to compare their findings in order to gauge the relative merits of treat-to-target and fire-and-forget.
Atorvastatin therapy in ASCOT (Table 2) reduced the risk of coronary death and nonfatal myocardial infarct by 36% (the absolute risk fell from 3% [154 events] on placebo to 1.9% [100 events] on statin,P=0.0005). Similarly, there were 121 stroke events (2.4%) in the placebo group and 90 (1.7%) in those prescribed the statin (P=0.024). Over the 3.3 year treatment period there were 397 deaths in total, but the 13% reduction in the statin-treated group was not statistically significant.
In order to compare ASCOT with WOSCOPS, we calculated the calendar date on which the WOSCOPS subjects had a median follow-up of 3.3 years, and analysed the results as they stood on that day. For the survivors, this gave a minimum follow-up duration of 23 months and a maximum duration of 55 months. After a median follow-up of 3.3 years, pravastatin had reduced (see Table 2) the risk of the primary endpoint of definite coronary death plus nonfatal myocardial infarction by 33% (from 180 events [5.5%] on placebo to 123 [37%] on pravastatin [P=0.0006]). There were only 52 stroke events within this time frame, with no difference between the treatment arms (P=0.61). However, treatment with pravastatin reduced (seeTable 2) the risk of death from all causes from 85 events (2.6%) on placebo to 62 (1.9%) on pravastatin (P=0.049).
Comparison of the 3.3-year truncated WOSCOPS data with the findings of ASCOT-LLA shows that, despite major differences in the on-treatment cholesterol values in the two studies, the benefits of statin therapy (whether pravastatin or atorvastatin) in the two were essentially the same. The observed 33% decrease in risk of the primary endpoint in WOSCOPS after a median follow-up of 3.3 years did not differ meaningfully from the reduction (31%) seen at the end of the study.14 This is despite the fact that only 20% of the WOSCOPS cohort achieved their guideline cholesterol target, compared with the 70% or so who did so in ASCOT-LLA. Moreover, an average intention-to-treat cholesterol reduction of 1.0 mmol/L was maintained over the course of the WOSCOPS trial, although the actual reduction in cholesterol was slightly less than 1.0 mmol/L in the final study year. So, it seems unlikely that the 1.3 mmol/L cholesterol reduction that resulted in the 36% fall in risk of the primary endpoint in ASCOT-LLA could have increased to anything approaching 50% by the original termination point of the project.22 Such a result is well beyond the experience of all other statin trials, as evidenced by the retrospective analysis of the CARE and LIPID data truncated to 3.3 years and shown in Figure 4 (Pravastatin Pooling Project Investigators, personal communication).23
Treatment with atorvastatin in ASCOT-LLA reduced the risk of stroke, evaluated as a secondary endpoint, by 27% at 3.3 years (P=0.024). In the WOSCOPS trial, stroke risk was not significantly reduced. The ASCOT-LLA study had more statistical power to evaluate the effect of treatment on stroke risk, the event rate in the control group being 2.4% compared with 1.6% at 4.9 years in the WOSCOPS control group, reflecting the greater age and higher prevalence of hypertension in the ASCOT-LLA cohort. Interestingly, the risk of death from any cause, an additional secondary endpoint, was not significantly reduced in ASCOT-LLA (decreased by 13%), probably again reflecting lack of statistical power and the early termination (seeTable 2). This endpoint was significantly reduced in WOSCOPS, at 3.3 years (28% risk reduction; 95% confidence interval, 0 to 40;P=0.049).
Comparison of the data from WOSCOPS and ASCOT-LLA suggests that zealous attention to achieving lipid guideline targets may be less productive than we have supposed. Wider prescription of low-dose statin therapy could make better use of limited resources and, in times of cost constraint, offer a more relevant treatment strategy for the population as a whole.